Computer Vision Neural Network Project: 11/9/2020 Summary

 I have been late, but I think what I will do is take a look at the neural network code and see what I can do to make some modifications. I want to benefit my team at least somewhat, if not to a major extent, as I have my name in the research paper, though I am taking 16-17 credit hours with all CS/engineering courses.

 

First, I have to install Anaconda, by installing PyTorch. I do this by going to https://www.anaconda.com/products/individual/download-success and clicking on the windows x64 button. I then start the setup. users/_____/Anaconda3 is the file that this was in. It set up the package cache and subsequently started to install. 

The second thing I have to do is to create a virtual environment variable by setting a certain Python command, called conda create --prefix ./venv python=3.7. in the Anaconda command line. 

I then typed the following command to successfully install the virtual environment: 

$ conda activate C:\Users\_____\venv

Subsequently, I installed pyTorch by using this command: 

conda install pytorch torchvision torchaudio cudatoolkit=10.2 -c pytorch

and afterwards check to see if it installed correctly in the Anaconda CLI.

It takes a good amount of time to install pytorch, so I am going to read the midterm report to see what I can do. 

The input to all this is an HDR image. Such is as follows: 

And there are N light sources, assumed as 3 in this project. These parameters are represented by an 8x1 vector, and the output is represented by an 8N x 1 vector with the following components where N is the number of light sources:

The sources is trained through the environment map to estimate these specific components of this light source. This is trained through a Convolutional Neural Network.

2 Metrics are 

1. The time it takes to run through the network

and

2. The time it takes to render an object. 

We want to perform preprocessing and discard inputs. 

The next step is to train a convolutional neural network. 

But first, I am going to finish the paper